Skip to content
M↓ View as Markdown ↗

Connect a data source

A data connection lets Actian AI Analyst read tables in your database or data warehouse. Models, Metrics, and SemQL queries run live against it, and AI Analysts answer questions through those Models.

Availability

  • Who can do this: members with the Edit capability in the active Workspace. If Restrict connection management is on, you also need the Manage organization capability (see Who can add connections).
  • Plan: individual connectors can be left out of your organization's plan. Studio shows only the connectors your organization can use.
  • Deployment: Actian SaaS and self-managed deployments. A self-managed deployment can offer fewer connectors.

Where connections live

In Studio, open Connections in the sidebar. The list groups your connections as Warehouses, Files, Catalog, and Channels. You see only the connections attached to your current Workspace. For details, see Data connections and Workspaces.

Click + (Add connection) to open the Add a connection page. It has four sections:

Section What it connects Where to read more
Warehouses Databases and data warehouses that AI Analyst queries live ("Query tables live for models, metrics, and SemQL.") This page and the connector pages below
Files A CSV file you upload and query like any other source CSV files
Catalog Actian Data Intelligence Platform, to sync business terms into the Glossary Catalog Connections and Actian Data Intelligence Platform
Channels Slack and Microsoft Teams, so people can use your AI Analysts there Messaging apps

Who can add connections

  • You need the Edit capability in the active Workspace to add, edit, or delete connections. Without it, the tiles on Add a connection are disabled.
  • A member with Manage organization can turn on Restrict connection management under Settings → Security (in the Administration group). While it is on, only members who also hold Manage organization can add, edit, or delete data connections in Studio. This includes uploading CSV files and adding catalog connections. Everyone else keeps read access and sees "Data connections are managed by your organization's Admins."

Supported warehouses

Every connector below needs a warehouse user with read access to the tables you want to use. The setup guides list the grants and network settings each one needs.

Connector (as shown in Studio) Setup guide Paste a connection string
PostgreSQL PostgreSQL Yes
MySQL MySQL Yes
MariaDB MariaDB Yes
Amazon Redshift AWS Redshift Yes
Microsoft SQL Server Microsoft SQL Server Yes
Azure Synapse Azure Synapse Yes
Microsoft Fabric Microsoft Fabric Yes
ClickHouse ClickHouse Yes
BigQuery BigQuery No
Snowflake Snowflake No
Databricks Databricks Yes
MotherDuck MotherDuck Yes
Actian Actian Data Platform Yes
Ingres Actian Ingres Yes
Informix Informix Yes
ZenDB Actian Zen Yes

How strongly each connector enforces read-only access differs. Some can't enforce it at all and rely on the warehouse user you supply. See Read-only access and query limits.

If a connector is missing

Add a connection lists only the connectors that your deployment offers and that your organization's plan includes. If a connector you need isn't listed, ask your administrator. If a connector is turned off later, existing connections of that type keep working: you can still browse, re-test, rename, and delete them. You can't create new ones or save new credentials for them.

Add a warehouse connection

Before you start, make sure that:

Then:

  1. In Studio, open Connections and click + (Add connection).
  2. Under Warehouses, choose your connector. The form opens with the connector's name as its title.
  3. Under Connection details, enter a Name and, if you like, a Description (optional).
  4. Leave Private to this Workspace unticked to create an organization connection, or tick it to keep the connection in this Workspace only. See Data connections and Workspaces.
  5. Under Credentials, fill in the connector's fields. To save typing, click Paste a connection string (if offered), paste the string, and click Fill the form. The fields are filled in your browser. Nothing is sent until you test the connection.
  6. Click Test connection. Wait for Connection successful.

  7. Pick what to expose:

    • Most connectors show a Schemas picker ("Select schemas to expose for browsing and model authoring"), filled with the schemas the warehouse user can see. The first schema is preselected. The selection is optional.
    • Databricks shows a Catalogs picker instead. Leave it empty to list every catalog the connection can see.
    • BigQuery (you enter a Dataset ID) and ZenDB show no picker.

    When you tick or untick a schema or catalog, the picker closes and the test result is cleared. Click Test connection again to bring the picker back with your selection kept, then save. 8. Click Save connection. Studio opens the new connection's page.

Things to know about the form:

  • Save connection stays disabled until the connection has a name and a successful test ("Test the connection before saving").
  • A successful test is valid for 5 minutes. If you save later than that, the save fails with "Connection test expired; test credentials again". Click Test connection again, then save.
  • Changing any credential field, including through Paste a connection string, clears the test result, so you test again. Changing Name, Description (optional), or Private to this Workspace doesn't.
  • For connectors that can't enforce read-only access themselves, the form shows "This Connector cannot enforce read-only access itself. Supply a warehouse user that is limited to read-only privileges."
  • Connection names must be unique. An organization connection's name must be unique across the organization, and a private connection's name must be unique within its Workspace. When you save a new connection with a name that is already taken, Studio refuses it with a message such as "A data source named '' already exists". If you rename a saved connection to a taken name, Studio shows "A connection with this name already exists".

Setup help panel

Next to the form, the Setup help panel has:

  • Before you connect: what the database must accept. If your deployment publishes the IP addresses it connects from, they are listed under Which IPs to allow.
  • Credentials: how credentials are stored (see below).
  • Need help from IT?: with Invite a teammate, which opens Settings → Members so you can invite the person who manages your firewall or database users.
  • Read the setup guide: a link to the connector's page in these docs.

How credentials are stored

Secret values, such as passwords, access tokens, private keys, client secrets, and service-account JSON, are encrypted at rest and never shown again after you save. Non-secret settings, such as host, port, database, username, and the selected schemas, appear on the connection's Overview tab.

Local and private addresses

If the Host looks like a local address (for example localhost, a 10.x or 192.168.x address, or a name without a domain), Studio shows This looks like a local address. The warning doesn't block the form, but you can only save after a successful test. AI Analyst connects from its own servers, not from your computer, so the test fails unless those servers can reach that address. See Network access and IP allowlisting.

If the test fails

Message What it usually means What to try
Authentication failed The warehouse rejected the username, password, token, or key. Check the credentials and the authentication settings for the connector.
Could not establish a secure connection The TLS/SSL handshake failed. Check the connector's SSL or encryption setting against what the server requires.
Could not reach the warehouse AI Analyst couldn't open a network connection. Check host and port, and allow AI Analyst through your firewall. See Network access and IP allowlisting.
This warehouse user cannot see those objects The user signed in but lacks access to the database, schema, or tables. Grant read access as described in the connector's setup guide.
This warehouse type isn't available right now This deployment isn't set up to reach this connector type. Contact your administrator.
Stored credentials could not be read The saved credentials for this connection can't be used. Re-enter them with Edit credentials (see Manage a data connection).
Could not connect Any other error. This includes a host name that doesn't resolve and a host that AI Analyst's network rules don't allow. Check the host value (hostname only, no https://, path, or port) and see Network access and IP allowlisting.

If required fields are empty, Studio asks you to fill them in before it tests ("Fill in … before testing the connection").

Next steps